I am currently reading paper on Variational Autoencoders by Dr. Kingma and Dr. Welling, and am having a few confusions.

enter image description here

Question 1: What is a spherical Gaussian latent space? How do we interpret it

Question 2: In equation 1.15, does the Roman I in standard deviation mean 1 or does it have a different meaning?

Could anyone please provide me some information on these?


1 Answer 1

  1. A spherical gaussian distribution is a multivariate normal distribution that has diagonal covariance and equal variances in all directions. A "spherical gaussian latent space" means that the latent space is a random variable with a spherical gaussian distribution.

  2. $I$ is the identity matrix.

  • $\begingroup$ Thank you @Sycorax! I have added a confusion in the description. It would be great if you could please help me understand it. $\endgroup$
    – Curious
    May 11, 2022 at 12:26
  • 1
    $\begingroup$ Sycorax answered the question you posted. If you have another, please post it as a separate question instead of editing it into the previous one. If you take advantage of the opportunity to do that, I encourage you to specify where the newly added figures are taken from. They’re not comparable to each other. $\endgroup$ May 11, 2022 at 12:44
  • $\begingroup$ @Curious Arya is correct — if you have a follow-up question, you should ask a new question $\endgroup$
    – Sycorax
    May 11, 2022 at 12:45
  • $\begingroup$ Sure, will add a new question. Thanks! Edit removed. $\endgroup$
    – Curious
    May 11, 2022 at 12:52

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.